Abstract
A 63-station radiosonde network has been used for many years to estimate temperature variations and trends at the surface and in the 850–300-, 300–100-, and 100–50-mb layers of climate zones, both hemispheres, and the globe, but with little regard for the quality of individual station data. In this paper, nine tropical radiosonde stations in this network are identified as anomalous based on unrepresentatively large standard-error-of-regression values for 300–100-mb trends for the period 1958–2000. In the Tropics the exclusion of the 9 anomalous stations from the 63-station network for 1958–2000 results in a warming of the 300–100-mb layer rather than a cooling, a doubling of the warming of the 850–300-mb layer to a value of 0.13 K decade−1, and a greater warming at 850–300-mb than at the surface. The global changes in trend are smaller, but include a change to the same warming of the surface and the 850–300-mb layer during 1958–2000. The effect of the station exclusions is much less for 1979–2000, suggesting that most of the data problems are before this time. Temperature trends based on the 63-station network are compared with the Microwave Sounding Unit (MSU) and other radiosonde trends, and agreement is better after the exclusion of the anomalous stations. There is consensus that in the Tropics the troposphere has warmed slightly more than the surface during 1958–2000, but that there has been a warming of the surface relative to the troposphere during 1979–2000. Globally, the warming of the surface and the troposphere are essentially the same during 1958–2000, but during 1979–2000 the surface warms more than the troposphere. During the latter period the radiosondes indicate considerably more low-stratospheric cooling in the Tropics than does the MSU.
Effect of Exclusion of Anomalous Tropical Stations on Temperature Trends from a 63-Station Radiosonde Network, and Comparison with Other Analyses